Preprints
https://doi.org/10.5194/egusphere-2024-1506
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/egusphere-2024-1506
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
11 Jun 2024
| 11 Jun 2024
Temporal dynamic vulnerability – Impact of antecedent events on residential building losses to wind storm events in Germany
Abstract. Severe winter storm events are one of Central Europe's most damaging natural hazards, therefore particularly in focus for disaster risk management. One key factor for risk is vulnerability. Risk assessments often assume vulnerability as constant. This is, however, not always a justifiable assumption. This work seeks and quantifies a potential dynamic of vulnerability for residential buildings in Germany. A likely factor affecting the dynamics of vulnerability is the hazard itself. As an extreme events may destroy the most vulnerable elements, it is likely that the subsequent rebuilding or repair will reduce their vulnerability for following events. Therefore, the intensity of the previous events and the resulting damage can be assumed to be a decisive factor in changing vulnerability. A second important factor is the time period between the previous and current event. If the next event occurs during the reconstruction phase, vulnerability might be higher than when the reconstruction phase is completed.
We analyze the role of previous storm events for the vulnerability of residential buildings. For this purpose, generalized additive models are implemented to estimate vulnerability as a function of the intensity of the previous event and the time interval between the events. The damage is extracted from a 23-year-long data set of the daily storm and hail losses for insured residential buildings in Germany on the administrative district level provided by the German Insurance Association, and the hazard component is described by the daily maximum wind load calculated from the ERA5 reanalysis. The results show a negative relationship between the previous event's intensity and the current event's damage. The duration between two events shows a significant reduction of the damage for events occurring one or more winter seasons ago compared to events occurring within the same season. On a daily scale, the first five to ten days are especially crucial for vulnerability reduction.
How to cite. Trojand, A., Rust, H., and Ulbrich, U.: Temporal dynamic vulnerability – Impact of antecedent events on residential building losses to wind storm events in Germany, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2024-1506, 2024.
Competing interests: One of the co-author is a member of the editorial board of Natural Hazards and Earth System Sciences.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.Download & links
Download & links
- Preprint
(1646 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor
| : Report abuse
-
RC1: 'Comment on egusphere-2024-1506', Tristian Stolte, 22 Aug 2024
General comments
This manuscript presents a new approach in which the effects of time and antecedent impact on building vulnerability is measured. The authors have shown that vulnerability to winter storms is influenced by the timing of previous events, the wind load of those events, and the previous loss ratios. It is an important contribution to the field of dynamic vulnerability and the science behind it appears to be solid. The authors managed to collect data which is often scarce for these type of assessments. Also, the Figures are of good quality and the combination of different variables is cleverly made. However, the conveyance of the science to the reader needs major revisions – especially in the discussion section, which is currently a collection of texts that belong in either the methods or results sections. The article also needs several technical corrections, for which I would suggest to let a native English speaker proofread the text.
The manuscript has great potential, and with major revisions on the text (mostly on how information is conveyed, but also technically), this could turn into an excellent article.
I should note that I am not an expert in General Additive Models, so I could not review this part of the methodology in detail.
Specific comments
- Introduction
- l21 & 23: Natural disasters or natural catastrophes do not exist. A disaster only is a disaster when humans or human objects are involved. Alternatively, the authors could use terms like “natural hazards”, “disasters from natural hazards”, or simply “disasters” (https://www.undrr.org/our-impact/campaigns/no-natural-disasters).
- I am glad to see that vulnerability is explicitly defined by the authors (l40-41). I did expect it earlier, where the authors first refer to the risk framework of the IPCC and UNDRR (l27-28). This may be a personal preference though, but I would like to ask the authors to reconsider the timing of presenting the reader with the definition of their main topic.
- The two sentences below sound counterintuitive. Do the authors mean that buildings experience moderate damage, but that the accumulated damage is nonetheless large? Please specify.
- “Although individual natural disasters, such as the flash flood that occurred in western Germany in July 2021 or the Elbe flood in 2002, are the most devastating single events, the accumulated damage from storm events caused three times more damage to residential buildings than other natural disasters between the years 2002 and 2021 (GDV, 2023).” (l20-24)
- “The impact on human life is relatively small, and the damage to individual buildings is generally moderate.” (l24-25)
- L49-53: the authors mention several building-related drivers of vulnerability, which they group under ‘non -hazard specific dynamics; . Then they refer to Drakes & Tate (2022) and Simpson et al. (2021), which are both on social vulnerability. And which do not have a clear link to buildings or the way in which non-hazard specific dynamics are portrayed here. Maybe the authors can try to solidify this bridge.
- L71-72: “The time between two events is crucial for winter storm events in Germany as these occur in a short amount of time more often than other hazards.” -> This sentence needs a reference. If Mühr et al. (2022) compare the frequency of winter storms with other hazards, then please make it explicit.
- Data:
- Depending on the journal’s requirements, the authors may be able to merge the Data and Methods sections, as there is no clear distinction between the two in terms of topic. I would say that the data processing is also part of the Methods, and that several data-related choices (like the pre-event threshold) are made in the Methods. It may also make more sense to combine sections 2.2 and 3.1 as I only understood why the authors added Table 1 when I arrived at section 3.1.
- L92-93: “…the damage is almost exclusively caused by windstorm events.” -> could the authors provide a source for this? Or is this in the insurance data?
- In the loss ratio (eq. 1), why would you multiply the insured sum by 103? If this is just the label from the insurance data, then I would suggest to leave the ‘103’ part out as it is a bit confusing form a mathematical standpoint.
- Could the authors include the formula for air density?
- Could the authors explain why it is important to include the “orography due to air density” (l108) into account?
- I do not follow the argumentation on why the authors buffer the districts with 31km buffers such that four ERA5 grid points can fit in, because the authors mention that the ERA5 also has a 31km resolution. A flowchart with some visuals can clearly demarcate such steps and help the reader to follow along with spatial data analysis. This also helps understanding the rest of the methods.
- Methods:
- In Figure 1 -> is an event one windstorm and the total loss ratio per district? Or is this defined differently? Please make this more explicit.
- If pre-event (as mentioned in the text) = previous event (as mentioned in Figure 1), then please make this explicit.
- I am not expert in GAMs, so it would be good to let someone else review this.
- L146-147: “Recent studies include only the hazard component and a parameter for exposure as covariates into the damage model approach
- (e.g. Heneka et al., 2006; Pardowitz et al., 2016; Welker et al., 2021).” Heneka et al. (2006) is not really a recent study anymore and Pardowitz et al. (2016) is also a bit on the edge of being recent. Maybe this can be rephrased to ‘past studies’?
- L157-158: “In model MDays the temporal scale is refined further to a diurnal time scale and only pre-events happened within four weeks (28 days) before the event are included in the model.” -> The authors seem to confuse the terms special scale with spatial resolution and use them interchangeably. The temporal scale is 28days, but the temporal resolution is diurnal, is that correct? See also: (l159-160) “These different temporal resolutions are chosen because shortly after the event occurrence, we expect the daily scale being essential…“.
- L159-160: “we expect the daily scale being essential and around one seventh of all previous events fall in this time span” -> why do the authors expect this and why one-seventh?
- The part on mean value 1914 was a bit hard to follow, although I understand the covariate’s purpose and the reasoning. What I don’t understand yet is what is the “number of contracts”. Contracts for what? Insurance contracts? And what does that mean for the m1914 value? Can the authors please make this more explicit.
- Results
- L176-177: “Especially events with wind loads larger 700N/m2 (Beaufort scale 12) and previous event loss ratios of more than 0.1‰ occur rarely.” Why do the authors highlight these and what is meant by ‘rarely’?
- L183-184: “Thus the vulnerability describing the loss ratio for a certain wind load is lower in cases with higher previous event loss ratios.” -> do the authors mean The loss ratio, which describes the vulnerability, for a certain wind load is lower in cases with higher previous event loss ratios?
- L196: “…lead to a significantly lower vulnerability for events with a pre-season event than with a same-season event…” -> where do the authors base this significance on? Is it statistically significant? Or do they mean something like ‘substantial’?
- Figures 7 and 9 could go to the appendix, maybe together with similar figures on the input data. The manuscript already has several interesting figures.
- Figure 10c shows an oddity that is not discussed: why is the 1.00 per mille line in between the 0.1 and 0.01 per mille lines for the first 4-5 days? The final paragraph of section 4.2.3 discusses Figures 10a and 10b but not 10c.
- Discussion
- L255-257: “Kreibich et al. (2023) focused on socio-hydrological data and used the Likert scale to estimate changes in the vulnerability, but, for example, did not quantify the impact of the time between two events.” This example comes a bit out of the blue and seems not very relevant for the argumentation.
- L270-274: “We consider only winter months for wind storm events and exclude summer months, as the damage data set includes storm and hail damages for residential buildings, but it is not possible to distinguish within the data between the causes for a certain loss. Similar to storm events, hail events mainly lead to roof and window damage of residential buildings. A hail event is likely to impact the vulnerability of residential buildings, which could not be included in this work due to the lack of hail data in the ERA5 data set and others. In general, extending the models to a multi hazard approach is desirable.” -> This whole part sits better in the methods as it presents reasons of why the authors investigate winter storms and not summer storms.
- L 275-277: “On a seasonal scale, the differences between same-season and pre-season pre-events increase strongly with exceeding a wind load of around 600N/m2 and thereby considerable structural damage according to the Beaufort scale. While on lower wind loads, only slight structural damages occur.” -> this is a repetition of the results.
- Overall, the discussion needs to be restructured. It is currently a collection of small paragraphs where the authors repeat results or briefly validate some methodological choices. Instead, it would be much more interesting to read about the implications of the results (which are very interesting!), like in the paragraph on l278 – 283. Also, more concrete advice on methods for follow-up studies could be presented. They could also talk about other factors that are important in (dynamic) vulnerability assessments, which are not included in this study. Another suggestion would be to talk more about the transferability of these results. And what about spatial differences between the regions under study?
Technical corrections
Overall, sentence structure is sometimes off. I would advice to let a native English speaker review the sentence structure to ensure that no unnecessary mistakes are made. Below are some examples, but not all of them are listed here.
- Introduction:
- Although = Not only (l20-21)
- The time between two events is crucial for winter storm events in Germany as these occur in a short amount of time and more often than other hazards. (l71-72)
- Data:
- L101: The ERA5 data has -> the ERA5 data have
- (e.g. … -> (e.g., …
- Methods
- L154… distinguishing a seasonal, weekly and a diurnal time scale
- L163: models results = models’ results? Model’s results? Something else?
- L165: “Therefore we use with the mean value 1914”
- L176: “Especially events with wind loads larger than 700N/m2”
- Results
- L180: “In Figure 3, the results for three different for the model MpreEvent.” -> sentence seems unfinished.
- l182: “…but also on the loss ratio occurring in the previous previous”
- Bluish -> blue (fig 5)
- Reddish -> red (fig 5)
- L229-230: “In case of a storm event”
- Discussion
- L282: “The more damage in the pre-event occurred…”.
- L284: “Additionally on daily and on weekly scale the vulnerability nearly stays constant…” -> on the daily and weekly scales. Or, on daily and weekly scales.
Citation: https://doi.org/10.5194/egusphere-2024-1506-RC1 -
AC2: 'Reply on RC1', Andreas Trojand, 06 Feb 2025
Thank you for carefully reading the manuscript and for your very positive assessment of our work's potential! We very much appreciate your detailled comments and helpful questions and we will revise the manuscript accordingly. For language editing, we will involve a native speaker for the revised version.
Regarding the specific comments:
Introduction
- l21 & 23: Natural disasters or natural catastrophes do not exist. A disaster only is a disaster when humans or human objects are involved. Alternatively, the authors could use terms like “natural hazards”, “disasters from natural hazards”, or simply “disasters” (https://www.undrr.org/our-impact/campaigns/no-natural-disasters).
- This is a good point! We change the concepts/wording accordingly.
- I am glad to see that vulnerability is explicitly defined by the authors (l40-41). I did expect it earlier, where the authors first refer to the risk framework of the IPCC and UNDRR (l27-28). This may be a personal preference though, but I would like to ask the authors to reconsider the timing of presenting the reader with the definition of their main topic.
- Thank you, we reconsider the timing for introducing the concept of vulnerability.
- The two sentences below sound counterintuitive. Do the authors mean that buildings experience moderate damage, but that the accumulated damage is nonetheless large? Please specify. 1. “Although individual natural disasters, such as the flash flood that occurred in western Germany in July 2021 or the Elbe flood in 2002, are the most devastating single events, the accumulated damage from storm events caused three times more damage to residential buildings than other natural disasters between the years 2002 and 2021 (GDV, 2023).” (l20-24) 2. “The impact on human life is relatively small, and the damage to individual buildings is generally moderate.” (l24-25)
- Thanks for pointing this out. We adress this issue.
- L49-53: the authors mention several building-related drivers of vulnerability, which they group under ‘non -hazard specific dynamics; . Then they refer to Drakes & Tate (2022) and Simpson et al. (2021), which are both on social vulnerability. And which do not have a clear link to buildings or the way in which non-hazard specific dynamics are portrayed here. Maybe the authors can try to solidify this bridge.
- Yes, this seems to be a valid point which we need to rethink and revise.
- L71-72: “The time between two events is crucial for winter storm events in Germany as these occur in a short amount of time more often than other hazards.” -> This sentence needs a reference. If Mühr et al. (2022) compare the frequency of winter storms with other hazards, then please make it explicit.
- Thanks! In the revised version, we give a clear link to a reference.
Data
- Depending on the journal’s requirements, the authors may be able to merge the Data and Methods sections, as there is no clear distinction between the two in terms of topic. I would say that the data processing is also part of the Methods, and that several data-related choices (like the pre-event threshold) are made in the Methods. It may also make more sense to combine sections 2.2 and 3.1 as I only understood why the authors added Table 1 when I arrived at section 3.1.
- Thanks for pointing us towards this irriatation. We consider merging section 2 and 3 for the revised version, or at least have no break between 2.2 and 3.1.
- L92-93: “…the damage is almost exclusively caused by windstorm events.” -> could the authors provide a source for this? Or is this in the insurance data?
- The insurance data covers losses related to contracs covering hail and storm. It is, however, not possible to derive the cause from the insurance data. A popular approximation for separating lossed due to hail and due to storm ios to separate the data set into a summer and a winter data set as hail is related to thunderstorms, a hazard predominantly occuring in summer and rarely in winter; winter storms are a winter phenomenon. We will provide a corresponding source in the review.
- In the loss ratio (eq. 1), why would you multiply the insured sum by 103? If this is just the label from the insurance data, then I would suggest to leave the 103 part out as it is a bit
confusing form a mathematical standpoint.- Yes, indeed, this is not a clear way of presenting the equation. The loss ratio is, however, typically given in euros loss per thousand euros insured sum. We will change the equation to a mathematical sound one.
- Could the authors include the formula for air density?
- The denstiy of air is either approximated via the ideal gas law of dry air or as the density of a mixture of two ideal gases: dry air and
water vapour. Both partial densities can be obtained from the ideal gas law. We will give a reference in the revised version.
- The denstiy of air is either approximated via the ideal gas law of dry air or as the density of a mixture of two ideal gases: dry air and
- Could the authors explain why it is important to include th “orography due to air density” (l108) into account?
- Thanks for pointing this out! We will change the sentence to "The wind load was chosen over the wind gust, as the wind load also takes the changing air density due to pressure changes with altitute, i.e. the orography into account,"
- I do not follow the argumentation on why the authors buffer the districts with 31km buffers such that four ERA5 grid points can fit in, because the authors mention that the ERA5 also has a 31km resolution. A flowchart with some visuals can clearly demarcate such steps and help the reader to follow along with spatial data analysis. This also helps understanding the rest of the methods.
- The procedure is a way to include grid points with centers just outside the districts but with their area reaching into the districts. We agree that this is not easy to understand from the actual description and a sketch/flow chart will help.
Methods
- In Figure 1 -> is an event one windstorm and the total loss ratio per district? Or is this defined differently? Please make this more explicit.
- For the event definition, we follow the thresholding approach of the German homeowner insurance. For an event, this threshold has to be exceeded by the maximum wind speed over all grid points assigned to the district. The loss ratios is given per district. We will revise this description to make that more clear.
- If pre-event (as mentioned in the text) = previous event (as mentioned in Figure 1), then please make this explicit.
- Thanks! Point taken and we will use a consistent naming of concepts in the revised version.
- I am not expert in GAMs, so it would be good to let someone else review this.
- We appreciate your honesty! A second reviewer commented on that section.
- L146-147: “Recent studies include only the hazard component and a parameter for exposure as covariates into the damage model approach (e.g. Heneka et al., 2006; Pardowitz et al., 2016; Welker et al., 2021).” Heneka et al. (2006) is not really a recent study anymore and Pardowitz et al. (2016) is also a bit on the edge of being recent. Maybe this can be rephrased to ‘past studies’?
- Will be changed accordingly. Thanks!
- L157-158: “In model MDays the temporal scale is refined further to a diurnal time scale and only pre-events happened within four weeks (28 days) before the event are included in the model.” -> The authors seem to confuse the terms special scale with spatial resolution and use them interchangeably. The temporal scale is 28days, but the temporal resolution is diurnal, is that correct? See also: (l159-160) “These different temporal resolutions are chosen because shortly after the event occurrence, we expect the daily scale being essential…“.
- We are not sure if we understand your question. The lines 157-158 do not mention any spatial scales. Also, we do not change the temporal resolution of the data (e.g. by aggregating); we do, however, create covariates which address previous events on time scales of either days (M_Days) or weeks (M_Week) or seasons (M_season). We realize, however, that this is not easy to understand and we will consider to rewrite this paragraph.
- L159-160: “we expect the daily scale being essential and around one seventh of all previous events fall in this time span” -> why do the authors expect this and why one-seventh?
- Thanks, we give more explanation in the revised version.
- The part on mean value 1914 was a bit hard to follow, although I understand the covariate’s purpose and the reasoning. What I don’t understand yet is what is the “number of contracts”. Contracts for what? Insurance contracts? And what does that mean for the m1914 value? Can the authors please make this more explicit.
- This is indeed a very difficult part and we agree that the description needs another iteration. Yes, we are talking about insurance contracts and as values measured in € change e.g. due to inflation, the value 1914 is a reference used to compare values (and losses) across decades. We give the description another iteration, thanks.
Results
- L176-177: “Especially events with wind loads larger 700N/m2 (Beaufort scale 12) and previous event loss ratios of more than 0.1‰ occur rarely.” Why do the authors highlight these and what is meant by ‘rarely’?
- We want to point out that there is a range of wind loads and losses which is very interesting for insurances but where data is sparse. The sentence needs more concrete information, like what is the ration of events over these thresholds. We give that in the revised version.
- L183-184: “Thus the vulnerability describing the loss ratio for a certain wind load is lower in cases with higher previous event loss ratios.” -> do the authors mean The loss ratio, which describes the vulnerability, for a certain wind load is lower in cases with higher previous event loss ratios?
- Thank you for pointing this out. Our main focus here is on vulnerability and not on the loss ratio. However, the wording is ambiguous and we will adjust it in the revision.
- L196: “…lead to a significantly lower vulnerability for events with a pre-season event than with a same-season event…” -> where do the authors base this significance on? Is it statistically significant? Or do they mean something like ‘substantial’?
- In this study, significantly different always means based on statistical significance. The shaded areas depict 95% confidence for the estimated of the expected loss ratios. If confidence intervals for "same-season pre events" and "pre-season pre-events" do not overlap, we can safely consider the difference as statistically significant. In the revised version, we add text to clarify this.
- Figures 7 and 9 could go to the appendix, maybe together with similar figures on the input data. The manuscript already has several interesting figures.
- Thanks for the advice and the compliment for the other figures. We will take this into account in the review.
- Figure 10c shows an oddity that is not discussed: why is the 1.00 per mille line in between the 0.1 and 0.01 per mille lines for the first 4-5 days? The final paragraph of section 4.2.3 discusses Figures 10a and 10b but not 10c.
- There are very few events where two such extreme events occur within 1-2 days of each other which leads to large uncertainties. We will add the missing description in the revised version.
Discussion
- L255-257: “Kreibich et al. (2023) focused on socio-hydrological data and used the Likert scale to estimate changes in the vulnerability, but, for example, did not quantify the impact of the time between two events.” This example comes a bit out of the blue and seems not very relevant for the argumentation.
- We agree that this example is not suitable here and will adapt it accordingly. Thanks
- L270-274: “We consider only winter months for wind storm events and exclude summer months, as the damage data set includes storm and hail damages for residential buildings, but it is not possible to distinguish within the data between the causes for a certain loss. Similar to storm events, hail events mainly lead to roof and window damage of residential buildings. A hail event is likely to impact the vulnerability of residential buildings, which could not be included in this work due to the lack of hail data in the ERA5 data set and others. In general, extending the models to a multi hazard approach is desirable.” -> This whole part sits better in the methods as it presents reasons of why the authors investigate winter storms and not summer storms.
- We consider moving this part to the earlier section on methods for the revised version.
- L275-277: “On a seasonal scale, the differences between same-season and pre-season pre-events increase strongly with exceeding a wind load of around 600N/m2 and thereby considerable structural damage according to the Beaufort scale. While on lower wind loads, only slight structural damages occur.” -> this is a repetition of the results.
- Thanks for the point, we remove this paragraph from the discussion.
- Overall, the discussion needs to be restructured. It is currently a collection of small paragraphs where the authors repeat results or briefly validate some methodological choices. Instead, it would be much more interesting to read about the implications of the results (which are very interesting!), like in the paragraph on l278 – 283. Also, more concrete advice on methods for follow-up studiescould be presented. They could also talk about other factors that are important in (dynamic) vulnerability assessments, which are not included in this study. Another suggestion would be to talk more about the transferability of these results. And what about spatial differences between the regions under study?
- Thank you for these points! They will surely help in restructuring the discussion section for the revised version.
Technical corrections
Overall, sentence structure is sometimes off. I would advice to let a
native English speaker review the sentence structure to ensure that no
unnecessary mistakes are made. Below are some examples, but not all of
them are listed here.- We will correct all points marked by the reviewer and involve a native
speaker in the language editing
Citation: https://doi.org/10.5194/egusphere-2024-1506-AC2 - l21 & 23: Natural disasters or natural catastrophes do not exist. A disaster only is a disaster when humans or human objects are involved. Alternatively, the authors could use terms like “natural hazards”, “disasters from natural hazards”, or simply “disasters” (https://www.undrr.org/our-impact/campaigns/no-natural-disasters).
- Introduction
-
RC2: 'Comment on egusphere-2024-1506', Anonymous Referee #2, 08 Oct 2024
The authors detect the role of previous storm events for the vulnerability of residential buildings in Germany based on 23-year-long daily storm losses data set and ERA 5 reanalysis data. This is an interesting work with a large workload. However, the manuscript lacks polish in its language, contains many minor grammatical errors, and the description of the methodology is not detailed enough. Therefore, I request a major revision. Some suggestions may be helpful are as follows:
- The claims in the Abstract shouldbe supported by data.
- Lines 29 –34, The content of this paragraph is not closely related to the topic of this article. It is recommended to adjust it to enhance its relevance to the title.
- The authors are advised to provide a summarizing analysis of the research gap at the end of the introduction and briefly introduce the research content.
- Lines 91 –92, If the influence of storms from other seasons is excluded, the authors need to explain whether the impacts of storms from other seasons are similar to those of winter storms.
- Line 127, There should be literature support here to justify the choice of 0.01%.
- Line 130, The symbols in the text are inconsistent with those in the formulas.
- Line 155, I don't quite understand the meaning of "binary information." Does it mean that we can only determine whether they occurred in the same winter season?
- Line 160, I also feel confused about “one seventh of all previous events fall in this time span”.
- Lines 179 –184, A significance analysis is needed here.
- The labels and legends in Figure 9 and Figure 10 are not clear.
Citation: https://doi.org/10.5194/egusphere-2024-1506-RC2 -
AC1: 'Reply on RC2', Andreas Trojand, 05 Feb 2025
Thank you for carefully reading the manuscript and for your positive assessment of the work's value! We very much appreciate your valuable suggestions. The revision will undergo language editing with the help of a native speaker to eliminate the language issues you pointed out.
Regarding your suggestions:
1. The claims in the Abstract should be supported by data.
- As far as points in the abstract are not self-evident, we will refer to literature or to the work presented in the manuscript.
2. Lines 29 –34, The content of this paragraph is not closely related to the topic of this article. It is recommended to adjust it to enhance its relevance to the title.
- In the mentioned paragraph in the introduction, we tried to put our research into a wider frame of storm impact research and pointed towards popular research activities (storm damage and changes in storm damage) to show that our research topic has not been covered extensively. We find that this is adequate but are happy to adjust this paragraph on request.
3. The authors are advised to provide a summarizing analysis of the research gap at the end of the introduction and briefly introduce the research content.
- Thank you for this point. That is definitively missing and will be added in the revision!
4. Lines 91 –92, If the influence of storms from other seasons is excluded, the authors need to explain whether the impacts of storms from other seasons are similar to those of winter storms.
- Unfortunately, we have not well formulated that case. The other seasons are not exclulded because they are similar but due to the ambiguity of the damage data in other seasons. The data does not contain information about the cause of the damage, it can be wind or hail. Only for the (extended) winter season, we can be relatively sure that the damage is caused by wind and not by hail. We will reformulate this part.
5. Line 127, There should be literature support here to justify the choice of 0.01%.
- Thank you for this point! We will motivate the choice of this threshold in the revised manuscript.
6. Line 130, The symbols in the text are inconsistent with those in the formulas.
- Thanks! We take care of this!
7. Line 155, I don't quite understand the meaning of "binary information." Does it mean that we can only determine whether they occurred in the same winter season?
- Good point! That is not clearly formulated. We introduce a dichotomous (binary) variable which can take take only two values, either "same season" or "previous season". We clarify this in the revised document.
8. Line 160, I also feel confused about “one seventh of all previous events fall in this time span”.
- Thanks! There is a problem with the sentence. We will fix this in the revision.
9. Lines 179 –184, A significance analysis is needed here.
- This part describes the results shown in Fig. 3 which shows the estimates of the expected loss ratio as a function of previous event loss ratio conditional on various wind loads. The point to be taken from this plot is that over the range of previous events' loss ratios, the actual events' loss ratio changed. The shaded areas depict 95% confidence for the estimated of the expected loss ratios. If confidence intervals for two different previous events' loss ratios do not overlap, we can safely consider the difference as statistically significant. In the revised version, we add text to clarify this.
10. The labels and legends in Figure 9 and Figure 10 are not clear.
- Thanks! In the revised version, we adjust these to become clearer.
Citation: https://doi.org/10.5194/egusphere-2024-1506-AC1
Viewed
Total article views: 491 (including HTML, PDF, and XML)
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 263 | 105 | 123 | 491 | 36 | 35 |
- HTML: 263
- PDF: 105
- XML: 123
- Total: 491
- BibTeX: 36
- EndNote: 35
Cumulative views and downloads
(calculated since 11 Jun 2024)
Viewed (geographical distribution)
Total article views: 484 (including HTML, PDF, and XML)
Thereof 484 with geography defined
and 0 with unknown origin.
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 169 | 34 |
| Germany | 2 | 68 | 14 |
| China | 3 | 35 | 7 |
| Netherlands | 4 | 35 | 7 |
| France | 5 | 14 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 169
1
169
Latest update: 06 Apr 2025
Andreas Trojand
CORRESPONDING AUTHOR
Institute for Meteorology, Freie Universität Berlin, Carl-Heinrich-Becker Weg 6-10, 12165 Berlin
Henning Rust
Institute for Meteorology, Freie Universität Berlin, Carl-Heinrich-Becker Weg 6-10, 12165 Berlin
Hans Ertel Centre for Weather Research, Carl-Heinrich-Becker Weg 6-10, 12165 Berlin
Uwe Ulbrich
Institute for Meteorology, Freie Universität Berlin, Carl-Heinrich-Becker Weg 6-10, 12165 Berlin
Short summary
The study investigates how the intensity of previous windstorm events and the time between two events affect the vulnerability of residential buildings in Germany. By analyzing 23 years of data, it was found that higher intensity of previous events generally reduces vulnerability in subsequent storms, while shorter intervals between events increase vulnerability. The results emphasize the approach of considering vulnerability in risk assessments as temporal dynamic.
The study investigates how the intensity of previous windstorm events and the time between two...